Commits
Wes McKinney authored d3cb6b47fde
ARROW-22: [C++] Convert flat Parquet schemas to Arrow schemas I'm going to limit the amount of nested data (especially repeated fields) cases in this patch as I haven't yet thought through the nested data reassembly from repetition / definition levels. Since the effective Arrow schemas may "collapse" multiple levels of nesting (for example: 3-level array encoding -- see https://github.com/apache/parquet-cpp/blob/master/src/parquet/schema/types.h), we'll need to track the logical correspondence between repetition and definition levels so that the right null bits can be set easily during reassembly. Closes #37. Closes #38. Closes #39 Author: Wes McKinney <wesm@apache.org> Author: Uwe L. Korn <uwelk@xhochy.com> Closes #41 from wesm/ARROW-22 and squashes the following commits: f388210 [Wes McKinney] Correct typo in Layout.md (thanks @takahirox) e5c429a [Wes McKinney] Test for some unsupported Parquet schema types, add unannotated FIXED_LEN_BYTE_ARRAY to List<UInt8> 54daa9b [Wes McKinney] Refactor tests to invoke FromParquetSchema 74d6bae [Wes McKinney] Convert BYTE_ARRAY to StringType or List<UInt8> depending on the logical type b7b9ca9 [Uwe L. Korn] Add basic conversion for primitive types 0e2a7f1 [Uwe L. Korn] Add macro for adding dependencies to tests 0dd1109 [Uwe L. Korn] ARROW-78: Add constructor for DecimalType